Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Extraction of the FuelBlockSecondaryKeyBlockHeights table to an off-chain worker #1671

Merged
merged 14 commits into from
Feb 22, 2024

Conversation

xgreenx
Copy link
Collaborator

@xgreenx xgreenx commented Feb 15, 2024

Closes #1583

The change moves the FuelBlockSecondaryKeyBlockHeights table and related logic to the off-chain worker. Along with this, the change adds a new Merklelized blueprint that maintains the binary Merkle tree over the storage data. It supports only the insertion of the objects without removing them. This blueprint replaces the logic that previously was defined in the Database.

Some side changes caused by the main change:

  • Now, each blueprint provides its own set of tests related to the logic of this blueprint.
  • The TransactionStatus uses block_height inside instead of the block_id.
  • The key for the dense merkle tree looks like:
    pub enum DenseMetadataKey<PrimaryKey> {
        /// The primary key of the `DenseMerkleMetadata`.
        Primary(PrimaryKey),
        #[default]
        /// The latest `DenseMerkleMetadata` of the table.
        Latest,
    }

@xgreenx xgreenx requested a review from a team February 15, 2024 15:52
@xgreenx xgreenx self-assigned this Feb 15, 2024
crates/fuel-core/Cargo.toml Outdated Show resolved Hide resolved
# Conflicts:
#	crates/fuel-core/src/database/storage.rs
#	crates/fuel-core/src/graphql_api/database.rs
#	crates/fuel-core/src/graphql_api/ports.rs
#	crates/fuel-core/src/graphql_api/storage.rs
#	crates/fuel-core/src/graphql_api/worker_service.rs
#	crates/storage/src/column.rs
@@ -745,9 +745,12 @@ impl FuelClient {
Ok(block)
}

pub async fn block_by_height(&self, height: u32) -> io::Result<Option<types::Block>> {
pub async fn block_by_height(
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The types on latest_gas_price include a u32 to represent block height. Should we use BlockHeight there?

/// The macro that generates basic storage tests for the table with the merklelized structure.
/// It uses the [`InMemoryStorage`](crate::structured_storage::test::InMemoryStorage).
#[macro_export]
macro_rules! basic_merklelized_storage_tests {
Copy link
Member

@MitchTurner MitchTurner Feb 17, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Should these macros be behind a test-helpers flag?

@@ -112,26 +112,31 @@ impl BlockHeader {
/// Set the block height for the header
pub fn set_block_height(&mut self, height: BlockHeight) {
self.consensus_mut().height = height;
self.recalculate_metadata();
Copy link
Member

@MitchTurner MitchTurner Feb 17, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It looks like we have this method in a separate impl block for some reason. In fact, we have multiple blocks for BlockHeader have the same (no) constraints. Can we combine the impl blocks?

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

They are segregated by their business logic. I'm okay with joining them into one/two impl blocks, but since they are in different places in the file, I prefer to not shuffle them in this PR to minimize number of changes=)

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Okay. I just got confused where a method was just reading the PR diffs, but with an IDE it doesn't really affect anything.

Comment on lines +138 to +144
pub enum DenseMetadataKey<PrimaryKey> {
/// The primary key of the `DenseMerkleMetadata`.
Primary(PrimaryKey),
#[default]
/// The latest `DenseMerkleMetadata` of the table.
Latest,
}
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I want to understand this better. If Latest gives you the entire latest Merkle tree, the PrimaryKey gives you a previous tree? Or a subsection?

Maybe this would be obvious if I understood the MTs terminology better.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes. The Merkle tree that we are using here is append-only. Because of that, we can only add new stuff at the end, and the Latest key allows us to do that. At the same time, appending a new leaf doesn't invalidate all previous Merkle trees, and PrimaryKey allows us to get the valid MerkleTree at some Key that we added before. For example, in the case of the FuelBlocks, we can get a valid Merkle tree at height X, X-1, .., X - Y and use it in proofs that some block is part of the Merkle tree at a specific height.

}

/// Set the previous root for the header
pub fn set_previous_root(&mut self, root: Bytes32) {
self.consensus_mut().prev_root = root;
self.recalculate_metadata();
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Should we use a builder in the hot paths instead of calling these setters one by one, now that they all have to recalculate the application header hash? We could save several hashing calls in the executor by calculating this after all the fields are set.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We use these methods only in the tests, so it doesn't hurt us=) I will put them under test-helper feature

@@ -180,7 +180,7 @@ pub enum TransactionStatus {
/// Transaction was successfully included in a block
Success {
/// Included in this block
block_id: BlockId,
block_height: BlockHeight,
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If there's a possibility of reorgs in the future, would only using block height on the api be problematic?

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

With a shared sequencer, I don't see how we can occur fork with several valid blocks for the same height. It is possible to have invalid blocks and valid blocks, but we will skip invalid due to invalidity and commit only a valid one.

Copy link
Member

@Voxelot Voxelot Feb 18, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

so if there are several proposed block ids for the same block height, and a user tries to see the block containing their tx by height, the tx might not exist in the block that actually gets accepted right?

Copy link
Collaborator Author

@xgreenx xgreenx Feb 18, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, the transaction has Committed status only when the corresponding block is finalized by the shared sequencer. Until finalisation point it is TxPool level and we can only provide soft confirmations without block information

$crate::paste::item! {
#[cfg(test)]
#[allow(unused_imports)]
mod [< $table:snake _basic_tests >] {
Copy link
Member

@Voxelot Voxelot Feb 18, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

These look identical to the basic storage tests, should we just reuse those here?

Also, perhaps we should have some merkle tree related tests as part of this macro? I.e similar to the root tests used on the sparse structures?

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The tests are not the same since the removal of elements is not allowed=)

I moved old tests for the FuelBlocks table in the blocks.rs instead of adding new tests here. I did it for the simplicity of the PR, but if you think it is better to make it part of this macro now, then later, I will do it=)

@xgreenx xgreenx requested a review from a team February 18, 2024 19:23
Copy link
Contributor

@bvrooman bvrooman left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Reviewed 22/47 files - will continue the review at a later point. A few questions/comments so far.

CHANGELOG.md Outdated Show resolved Hide resolved
crates/client/src/client/types.rs Outdated Show resolved Hide resolved
@@ -178,7 +178,7 @@ pub struct SubmittedStatus {
#[cynic(schema_path = "./assets/schema.sdl")]
pub struct SuccessStatus {
pub transaction_id: TransactionId,
pub block: BlockIdFragment,
pub block: BlockHeightFragment,
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nit: Maybe this should be named block_height now?

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It is named block since from the GraphQL perspective you receive the block=) The block_height is ID, but it could be BlockId as well(both works)

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Agreed - since this could return a full block, we must use a fragment query here to only return a partial view of the object.

/// It links the `BlockId` to corresponding `BlockHeight`.
pub struct FuelBlockSecondaryKeyBlockHeights;

impl Mappable for FuelBlockSecondaryKeyBlockHeights {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If I understand correctly, this simply maps BlockId to BlockHeight. The naming suggests that this is a secondary storage for BlockHeight, i.e. a second way to query BlockHeight. The metadata table is the first way, and metadata is on-chain.

Since BlockHeight is now off-chain, I don't think it is "secondary" - we have separate on-chain and off-chain domains. I think now this could just be called BlockHeights, which is more clear to the reader.

Copy link
Member

@Voxelot Voxelot Feb 20, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

maybe FuelBlockIdsToHeights? I also find it confusing to parse out from this table name what it's actually doing without having to inspect the blueprint.

/// See [`SealedBlockConsensus`](crate::tables::SealedBlockConsensus)
FuelBlockConsensus = 19,
FuelBlockConsensus = 18,
Copy link
Contributor

@bvrooman bvrooman Feb 19, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Will this be a breaking change for existing databases and require a migration? My understanding is that we have the manual numbering to make sure that there isn't any breakage when adding/removing columns. Otherwise, we could just use the default implicit numbering.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Since we don't have mainnet it's okay to break the numbering and invalidate the database. But yeah, later we will try to avoid breaking stuff like this

}

impl<D: OnChainDatabase + ?Sized> SimpleBlockData for D {
impl<D: OnChainDatabase + OffChainDatabase + ?Sized> SimpleBlockData for D {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It seems a bit odd that we are giving access to on-chain data to off-chain views. Shouldn't on-chain and off-chain databases be fully self-sufficient?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It looks like this constraint isn't even needed. I removed it from here and line 53 impl<D: OnChainDatabase + ?Sized> BlockQueryData for D { and the tests passed.

@xgreenx xgreenx requested a review from a team February 21, 2024 18:58
@xgreenx xgreenx merged commit ed92738 into master Feb 22, 2024
31 checks passed
@xgreenx xgreenx deleted the feature/blocks-second-key-extraction branch February 22, 2024 00:34
@xgreenx xgreenx mentioned this pull request Mar 4, 2024
xgreenx added a commit that referenced this pull request Mar 4, 2024
## Version v0.23.0

### Added

- [#1713](#1713): Added
automatic `impl` of traits `StorageWrite` and `StorageRead` for
`StructuredStorage`. Tables that use a `Blueprint` can be read and
written using these interfaces provided by structured storage types.
- [#1671](#1671): Added a new
`Merklized` blueprint that maintains the binary Merkle tree over the
storage data. It supports only the insertion of the objects without
removing them.
- [#1657](#1657): Moved
`ContractsInfo` table from `fuel-vm` to on-chain tables, and created
version-able `ContractsInfoType` to act as the table's data type.

### Changed

- [#1723](#1723): Notify about
imported blocks from the off-chain worker.
- [#1717](#1717): The fix for
the [#1657](#1657) to include
the contract into `ContractsInfo` table.
- [#1657](#1657): Upgrade to
`fuel-vm` 0.46.0.
- [#1671](#1671): The logic
related to the `FuelBlockIdsToHeights` is moved to the off-chain worker.
- [#1663](#1663): Reduce the
punishment criteria for mempool gossipping.
- [#1658](#1658): Removed
`Receipts` table. Instead, receipts are part of the
`TransactionStatuses` table.
- [#1640](#1640): Upgrade to
fuel-vm 0.45.0.
- [#1635](#1635): Move
updating of the owned messages and coins to off-chain worker.
- [#1650](#1650): Add api
endpoint for getting estimates for future gas prices
- [#1649](#1649): Add api
endpoint for getting latest gas price
- [#1600](#1640): Upgrade to
fuel-vm 0.45.0
- [#1633](#1633): Notify
services about importing of the genesis block.
- [#1625](#1625): Making
relayer independent from the executor and preparation for the force
transaction inclusion.
- [#1613](#1613): Add api
endpoint to retrieve a message by its nonce.
- [#1612](#1612): Use
`AtomicView` in all services for consistent results.
- [#1597](#1597): Unify
namespacing for `libp2p` modules
- [#1591](#1591): Simplify
libp2p dependencies and not depend on all sub modules directly.
- [#1590](#1590): Use
`AtomicView` in the `TxPool` to read the state of the database during
insertion of the transactions.
- [#1587](#1587): Use
`BlockHeight` as a primary key for the `FuelsBlock` table.
- [#1585](#1585): Let
`NetworkBehaviour` macro generate `FuelBehaviorEvent` in p2p
- [#1579](#1579): The change
extracts the off-chain-related logic from the executor and moves it to
the GraphQL off-chain worker. It creates two new concepts - Off-chain
and On-chain databases where the GraphQL worker has exclusive ownership
of the database and may modify it without intersecting with the On-chain
database.
- [#1577](#1577): Moved
insertion of sealed blocks into the `BlockImporter` instead of the
executor.
- [#1574](#1574): Penalizes
peers for sending invalid responses or for not replying at all.
- [#1601](#1601): Fix
formatting in docs and check that `cargo doc` passes in the CI.
- [#1636](#1636): Add more
docs to GraphQL DAP API.

#### Breaking

- [#1725](#1725): All API
endpoints now are prefixed with `/v1` version. New usage looks like:
`/v1/playground`, `/v1/graphql`, `/v1/graphql-sub`, `/v1/metrics`,
`/v1/health`.
- [#1722](#1722): Bugfix: Zero
`predicate_gas_used` field during validation of the produced block.
- [#1714](#1714): The change
bumps the `fuel-vm` to `0.47.1`. It breaks several breaking changes into
the protocol:
- All malleable fields are zero during the execution and unavailable
through the GTF getters. Accessing them via the memory directly is still
possible, but they are zero.
- The `Transaction` doesn't define the gas price anymore. The gas price
is defined by the block producer and recorded in the `Mint` transaction
at the end of the block. A price of future blocks can be fetched through
a [new API nedopoint](#1641)
and the price of the last block can be fetch or via the block or another
[API endpoint](#1647).
- The `GasPrice` policy is replaced with the `Tip` policy. The user may
specify in the native tokens how much he wants to pay the block producer
to include his transaction in the block. It is the prioritization
mechanism to incentivize the block producer to include users
transactions earlier.
- The `MaxFee` policy is mandatory to set. Without it, the transaction
pool will reject the transaction. Since the block producer defines the
gas price, the only way to control how much user agreed to pay can be
done only through this policy.
- The `maturity` field is removed from the `Input::Coin`. The same
affect can be achieve with the `Maturity` policy on the transaction and
predicate. This changes breaks how input coin is created and removes the
passing of this argument.
- The metadata of the `Checked<Tx>` doesn't contain `max_fee` and
`min_fee` anymore. Only `max_gas` and `min_gas`. The `max_fee` is
controlled by the user via the `MaxFee` policy.
- Added automatic `impl` of traits `StorageWrite` and `StorageRead` for
`StructuredStorage`. Tables that use a `Blueprint` can be read and
written using these interfaces provided by structured storage types.

- [#1712](#1712): Make
`ContractUtxoInfo` type a version-able enum for use in the
`ContractsLatestUtxo`table.
- [#1657](#1657): Changed
`CROO` gas price type from `Word` to `DependentGasPrice`. The dependent
gas price values are dummy values while awaiting updated benchmarks.
- [#1671](#1671): The GraphQL
API uses block height instead of the block id where it is possible. The
transaction status contains `block_height` instead of the `block_id`.
- [#1675](#1675): Simplify GQL
schema by disabling contract resolvers in most cases, and just return a
ContractId scalar instead.
- [#1658](#1658): Receipts are
part of the transaction status.
Removed `reason` from the `TransactionExecutionResult::Failed`. It can
be calculated based on the program state and receipts.
Also, it is not possible to fetch `receipts` from the `Transaction`
directly anymore. Instead, you need to fetch `status` and its receipts.
- [#1646](#1646): Remove
redundant receipts from queries.
- [#1639](#1639): Make Merkle
metadata, i.e. `SparseMerkleMetadata` and `DenseMerkleMetadata` type
version-able enums
- [#1632](#1632): Make
`Message` type a version-able enum
- [#1631](#1631): Modify api
endpoint to dry run multiple transactions.
- [#1629](#1629): Use a
separate database for each data domain. Each database has its own folder
where data is stored.
- [#1628](#1628): Make
`CompressedCoin` type a version-able enum
- [#1616](#1616): Make
`BlockHeader` type a version-able enum
- [#1614](#1614): Use the
default consensus key regardless of trigger mode. The change is breaking
because it removes the `--dev-keys` argument. If the `debug` flag is
set, the default consensus key will be used, regardless of the trigger
mode.
- [#1596](#1596): Make
`Consensus` type a version-able enum
- [#1593](#1593): Make `Block`
type a version-able enum
- [#1576](#1576): The change
moves the implementation of the storage traits for required tables from
`fuel-core` to `fuel-core-storage` crate. The change also adds a more
flexible configuration of the encoding/decoding per the table and allows
the implementation of specific behaviors for the table in a much easier
way. It unifies the encoding between database, SMTs, and iteration,
preventing mismatching bytes representation on the Rust type system
level. Plus, it increases the re-usage of the code by applying the same
blueprint to other tables.
    
It is a breaking PR because it changes database encoding/decoding for
some tables.
    
    ### StructuredStorage
    
The change adds a new type `StructuredStorage`. It is a wrapper around
the key-value storage that implements the storage
traits(`StorageInspect`, `StorageMutate`, `StorageRead`, etc) for the
tables with blueprint. This blueprint works in tandem with the
`TableWithBlueprint` trait. The table may implement `TableWithBlueprint`
specifying the blueprint, as an example:
    
    ```rust
    impl TableWithBlueprint for ContractsRawCode {
        type Blueprint = Plain<Raw, Raw>;
    
        fn column() -> Column {
            Column::ContractsRawCode
        }
    }
    ```
    
It is a definition of the blueprint for the `ContractsRawCode` table. It
has a plain blueprint meaning it simply encodes/decodes bytes and
stores/loads them into/from the storage. As a key codec and value codec,
it uses a `Raw` encoding/decoding that simplifies writing bytes and
loads them back into the memory without applying any serialization or
deserialization algorithm.
    
If the table implements `TableWithBlueprint` and the selected codec
satisfies all blueprint requirements, the corresponding storage traits
for that table are implemented on the `StructuredStorage` type.
    
    ### Codecs
    
Each blueprint allows customizing the key and value codecs. It allows
the use of different codecs for different tables, taking into account
the complexity and weight of the data and providing a way of more
optimal implementation.
    
That property may be very useful to perform migration in a more easier
way. Plus, it also can be a `no_std` migration potentially allowing its
fraud proving.
    
    An example of migration:
    
    ```rust
    /// Define the table for V1 value encoding/decoding.
    impl TableWithBlueprint for ContractsRawCodeV1 {
        type Blueprint = Plain<Raw, Raw>;
    
        fn column() -> Column {
            Column::ContractsRawCode
        }
    }
    
    /// Define the table for V2 value encoding/decoding.
    /// It uses `Postcard` codec for the value instead of `Raw` codec.
    ///
    /// # Dev-note: The columns is the same.
    impl TableWithBlueprint for ContractsRawCodeV2 {
        type Blueprint = Plain<Raw, Postcard>;
    
        fn column() -> Column {
            Column::ContractsRawCode
        }
    }
    
    fn migration(storage: &mut Database) {
        let mut iter = storage.iter_all::<ContractsRawCodeV1>(None);
        while let Ok((key, value)) = iter.next() {
            // Insert into the same table but with another codec.
            storage.storage::<ContractsRawCodeV2>().insert(key, value);
        }
    }
    ```
    
    ### Structures
    
The blueprint of the table defines its behavior. As an example, a
`Plain` blueprint simply encodes/decodes bytes and stores/loads them
into/from the storage. The `SMT` blueprint builds a sparse merkle tree
on top of the key-value pairs.
    
Implementing a blueprint one time, we can apply it to any table
satisfying the requirements of this blueprint. It increases the re-usage
of the code and minimizes duplication.
    
It can be useful if we decide to create global roots for all required
tables that are used in fraud proving.
    
    ```rust
    impl TableWithBlueprint for SpentMessages {
        type Blueprint = Plain<Raw, Postcard>;
    
        fn column() -> Column {
            Column::SpentMessages
        }
    }
                     |
                     |
                    \|/
    
    impl TableWithBlueprint for SpentMessages {
        type Blueprint =
Sparse<Raw, Postcard, SpentMessagesMerkleMetadata,
SpentMessagesMerkleNodes>;
    
        fn column() -> Column {
            Column::SpentMessages
        }
    }
    ```
    
    ### Side changes
    
    #### `iter_all`
The `iter_all` functionality now accepts the table instead of `K` and
`V` generics. It is done to use the correct codec during
deserialization. Also, the table definition provides the column.
    
    #### Duplicated unit tests
    
The `fuel-core-storage` crate provides macros that generate unit tests.
Almost all tables had the same test like `get`, `insert`, `remove`,
`exist`. All duplicated tests were moved to macros. The unique one still
stays at the same place where it was before.
    
    #### `StorageBatchMutate`
    
Added a new `StorageBatchMutate` trait that we can move to
`fuel-storage` crate later. It allows batch operations on the storage.
It may be more performant in some cases.

- [#1573](#1573): Remove
nested p2p request/response encoding. Only breaks p2p networking
compatibility with older fuel-core versions, but is otherwise fully
internal.

## What's Changed
* Fix the example tx in README.md by @Dentosal in
#1543
* Weekly `cargo update` by @github-actions in
#1557
* Update LICENSE by @Voxelot in
#1558
* Remove duplicating logic in the `KeyValueStore` trait by @xgreenx in
#1559
* Move `KeyValueStore` to the `fuel-core-storage` crate by @xgreenx in
#1566
* Update `libp2p` from `0.50.0` to `0.53.1` by @MitchTurner in
#1379
* Weekly `cargo update` by @github-actions in
#1575
* Weekly `cargo update` by @github-actions in
#1578
* Moved insertion of the blocks into the `BlockImporter` instead of the
executor by @xgreenx in #1577
* Let `NetworkBehaviour` macro generate `FuelBehaviorEvent` in p2p by
@MitchTurner in #1585
* Remove RustSec Ignore for RUSTSEC-2022-0093 by @MitchTurner in
#1586
* Simplify p2p request/response message serialization by @Dentosal in
#1573
* Weekly `cargo update` by @github-actions in
#1588
* Depend on libp2p directly instead of submodules by @MitchTurner in
#1591
* Make `Block` versionable by @MitchTurner in
#1593
* Switch katyo/publish-crates@v2 to xgreenx/publish-crates@v1 for the CI
by @Dentosal in #1603
* Weekly `cargo update` by @github-actions in
#1599
* Fix cargo doc, check for cargo doc in the CI by @Dentosal in
#1601
* Move storage traits implementation to the `fuel-core-storage` crate by
@xgreenx in #1576
* Extract off chain logic from the executor by @xgreenx in
#1579
* Applying comments from the PR for the storage crate by @xgreenx in
#1610
* Use `BlockHeight` as a primary key for the `FuelsBlock` table by
@xgreenx in #1587
* Use `AtomicView` in the `TxPool` by @xgreenx in
#1590
* Unify namespacing in P2P codebase by @MitchTurner in
#1597
* Weekly `cargo update` by @github-actions in
#1611
* Make `Consensus` version-able by @MitchTurner in
#1596
* feat: add api endpoint to retrieve messages by nonce by @matt-user in
#1613
* Update Docker tags for Flux integration by @liviudm in
#1617
* Remove dev consensus key by @MujkicA in
#1614
* Make `BlockHeader` a versionable enum by @MitchTurner in
#1616
* feat: Versionable CompressedCoin by @bvrooman in
#1628
* Use `AtomicView` in all services by @xgreenx in
#1612
* Notify services about importing of the genesis block by @xgreenx in
#1633
* Weekly `cargo update` by @github-actions in
#1634
* feat: Versionable `Message` by @bvrooman in
#1632
* Making relayer independent from the executor by @xgreenx in
#1625
* Upgrade to the fuel-vm 0.44.0 by @Dentosal in
#1600
* Upgrade to fuel-vm 0.45.0 by @MitchTurner in
#1640
* feat: Versionable Merkle metadata by @bvrooman in
#1639
* Add docs for GraphQL DAP endpoints by @Dentosal in
#1636
* Use a separate database for each data domain by @xgreenx in
#1629
* feat: dry run multiple transactions by @matt-user in
#1631
* Weekly `cargo update` by @github-actions in
#1644
* Decrease peer reputation on request timeouts and decode errors by
@Dentosal in #1574
* refactor: remove redudant receipts queries by @matt-user in
#1646
* Latest gas price endpoint by @MitchTurner in
#1649
* Estimate gas price API Endpoint by @MitchTurner in
#1650
* Fix for the race condition with tx status and receipts by @xgreenx in
#1658
* Bump up rust to 1.75 by @MitchTurner in
#1656
* Weekly `cargo update` by @github-actions in
#1655
* Duplication of the `0.22.1` release by @xgreenx in
#1665
* Move updating of the owned messages and coins to off-chain worker by
@xgreenx in #1635
* Weekly `cargo update` by @github-actions in
#1682
* Remove nested contract resolvers in GQL schema by @Voxelot in
#1675
* Soften txpool p2p reputation requirements by @Voxelot in
#1663
* Remove failing deployment job by @Voxelot in
#1692
* chore: format cargo tomls and add ci check by @segfault-magnet in
#1691
* Extraction of the `FuelBlockSecondaryKeyBlockHeights` table to an
off-chain worker by @xgreenx in
#1671
* Weekly `cargo update` by @github-actions in
#1707
* Increase liveness probe timeout to 5 minutes by @rfuelsh in
#1709
* feat: Add `ContractInfo` table and versionable `ContractInfoType` to
on-chain storage by @bvrooman in
#1657
* feat: Versionable `ContractUtxoInfo` for the `ContractsLatestUtxo`
table by @bvrooman in #1712
* Weekly `cargo update` by @github-actions in
#1719
* Bump `fuel-vm` dep to 0.47.1 by @MitchTurner in
#1714
* Bugfix: Calling of the contract with enabled `utxo_validation` fails
by @bvrooman in #1717
* Bugfix: Zero `predicate_gas_used` field during validation of the
produced block by @xgreenx in
#1722
* Notify about imported blocks from the off-chain worker by @xgreenx in
#1723
* Added versioning suffix to the GraphQL endpoint by @xgreenx in
#1725

## New Contributors
* @liviudm made their first contribution in
#1617

**Full Changelog**:
v0.22.0...v0.23.0
goldenpath1109 added a commit to goldenpath1109/fuel-core that referenced this pull request Sep 7, 2024
## Version v0.23.0

### Added

- [#1713](FuelLabs/fuel-core#1713): Added
automatic `impl` of traits `StorageWrite` and `StorageRead` for
`StructuredStorage`. Tables that use a `Blueprint` can be read and
written using these interfaces provided by structured storage types.
- [#1671](FuelLabs/fuel-core#1671): Added a new
`Merklized` blueprint that maintains the binary Merkle tree over the
storage data. It supports only the insertion of the objects without
removing them.
- [#1657](FuelLabs/fuel-core#1657): Moved
`ContractsInfo` table from `fuel-vm` to on-chain tables, and created
version-able `ContractsInfoType` to act as the table's data type.

### Changed

- [#1723](FuelLabs/fuel-core#1723): Notify about
imported blocks from the off-chain worker.
- [#1717](FuelLabs/fuel-core#1717): The fix for
the [#1657](FuelLabs/fuel-core#1657) to include
the contract into `ContractsInfo` table.
- [#1657](FuelLabs/fuel-core#1657): Upgrade to
`fuel-vm` 0.46.0.
- [#1671](FuelLabs/fuel-core#1671): The logic
related to the `FuelBlockIdsToHeights` is moved to the off-chain worker.
- [#1663](FuelLabs/fuel-core#1663): Reduce the
punishment criteria for mempool gossipping.
- [#1658](FuelLabs/fuel-core#1658): Removed
`Receipts` table. Instead, receipts are part of the
`TransactionStatuses` table.
- [#1640](FuelLabs/fuel-core#1640): Upgrade to
fuel-vm 0.45.0.
- [#1635](FuelLabs/fuel-core#1635): Move
updating of the owned messages and coins to off-chain worker.
- [#1650](FuelLabs/fuel-core#1650): Add api
endpoint for getting estimates for future gas prices
- [#1649](FuelLabs/fuel-core#1649): Add api
endpoint for getting latest gas price
- [#1600](FuelLabs/fuel-core#1640): Upgrade to
fuel-vm 0.45.0
- [#1633](FuelLabs/fuel-core#1633): Notify
services about importing of the genesis block.
- [#1625](FuelLabs/fuel-core#1625): Making
relayer independent from the executor and preparation for the force
transaction inclusion.
- [#1613](FuelLabs/fuel-core#1613): Add api
endpoint to retrieve a message by its nonce.
- [#1612](FuelLabs/fuel-core#1612): Use
`AtomicView` in all services for consistent results.
- [#1597](FuelLabs/fuel-core#1597): Unify
namespacing for `libp2p` modules
- [#1591](FuelLabs/fuel-core#1591): Simplify
libp2p dependencies and not depend on all sub modules directly.
- [#1590](FuelLabs/fuel-core#1590): Use
`AtomicView` in the `TxPool` to read the state of the database during
insertion of the transactions.
- [#1587](FuelLabs/fuel-core#1587): Use
`BlockHeight` as a primary key for the `FuelsBlock` table.
- [#1585](FuelLabs/fuel-core#1585): Let
`NetworkBehaviour` macro generate `FuelBehaviorEvent` in p2p
- [#1579](FuelLabs/fuel-core#1579): The change
extracts the off-chain-related logic from the executor and moves it to
the GraphQL off-chain worker. It creates two new concepts - Off-chain
and On-chain databases where the GraphQL worker has exclusive ownership
of the database and may modify it without intersecting with the On-chain
database.
- [#1577](FuelLabs/fuel-core#1577): Moved
insertion of sealed blocks into the `BlockImporter` instead of the
executor.
- [#1574](FuelLabs/fuel-core#1574): Penalizes
peers for sending invalid responses or for not replying at all.
- [#1601](FuelLabs/fuel-core#1601): Fix
formatting in docs and check that `cargo doc` passes in the CI.
- [#1636](FuelLabs/fuel-core#1636): Add more
docs to GraphQL DAP API.

#### Breaking

- [#1725](FuelLabs/fuel-core#1725): All API
endpoints now are prefixed with `/v1` version. New usage looks like:
`/v1/playground`, `/v1/graphql`, `/v1/graphql-sub`, `/v1/metrics`,
`/v1/health`.
- [#1722](FuelLabs/fuel-core#1722): Bugfix: Zero
`predicate_gas_used` field during validation of the produced block.
- [#1714](FuelLabs/fuel-core#1714): The change
bumps the `fuel-vm` to `0.47.1`. It breaks several breaking changes into
the protocol:
- All malleable fields are zero during the execution and unavailable
through the GTF getters. Accessing them via the memory directly is still
possible, but they are zero.
- The `Transaction` doesn't define the gas price anymore. The gas price
is defined by the block producer and recorded in the `Mint` transaction
at the end of the block. A price of future blocks can be fetched through
a [new API nedopoint](FuelLabs/fuel-core#1641)
and the price of the last block can be fetch or via the block or another
[API endpoint](FuelLabs/fuel-core#1647).
- The `GasPrice` policy is replaced with the `Tip` policy. The user may
specify in the native tokens how much he wants to pay the block producer
to include his transaction in the block. It is the prioritization
mechanism to incentivize the block producer to include users
transactions earlier.
- The `MaxFee` policy is mandatory to set. Without it, the transaction
pool will reject the transaction. Since the block producer defines the
gas price, the only way to control how much user agreed to pay can be
done only through this policy.
- The `maturity` field is removed from the `Input::Coin`. The same
affect can be achieve with the `Maturity` policy on the transaction and
predicate. This changes breaks how input coin is created and removes the
passing of this argument.
- The metadata of the `Checked<Tx>` doesn't contain `max_fee` and
`min_fee` anymore. Only `max_gas` and `min_gas`. The `max_fee` is
controlled by the user via the `MaxFee` policy.
- Added automatic `impl` of traits `StorageWrite` and `StorageRead` for
`StructuredStorage`. Tables that use a `Blueprint` can be read and
written using these interfaces provided by structured storage types.

- [#1712](FuelLabs/fuel-core#1712): Make
`ContractUtxoInfo` type a version-able enum for use in the
`ContractsLatestUtxo`table.
- [#1657](FuelLabs/fuel-core#1657): Changed
`CROO` gas price type from `Word` to `DependentGasPrice`. The dependent
gas price values are dummy values while awaiting updated benchmarks.
- [#1671](FuelLabs/fuel-core#1671): The GraphQL
API uses block height instead of the block id where it is possible. The
transaction status contains `block_height` instead of the `block_id`.
- [#1675](FuelLabs/fuel-core#1675): Simplify GQL
schema by disabling contract resolvers in most cases, and just return a
ContractId scalar instead.
- [#1658](FuelLabs/fuel-core#1658): Receipts are
part of the transaction status.
Removed `reason` from the `TransactionExecutionResult::Failed`. It can
be calculated based on the program state and receipts.
Also, it is not possible to fetch `receipts` from the `Transaction`
directly anymore. Instead, you need to fetch `status` and its receipts.
- [#1646](FuelLabs/fuel-core#1646): Remove
redundant receipts from queries.
- [#1639](FuelLabs/fuel-core#1639): Make Merkle
metadata, i.e. `SparseMerkleMetadata` and `DenseMerkleMetadata` type
version-able enums
- [#1632](FuelLabs/fuel-core#1632): Make
`Message` type a version-able enum
- [#1631](FuelLabs/fuel-core#1631): Modify api
endpoint to dry run multiple transactions.
- [#1629](FuelLabs/fuel-core#1629): Use a
separate database for each data domain. Each database has its own folder
where data is stored.
- [#1628](FuelLabs/fuel-core#1628): Make
`CompressedCoin` type a version-able enum
- [#1616](FuelLabs/fuel-core#1616): Make
`BlockHeader` type a version-able enum
- [#1614](FuelLabs/fuel-core#1614): Use the
default consensus key regardless of trigger mode. The change is breaking
because it removes the `--dev-keys` argument. If the `debug` flag is
set, the default consensus key will be used, regardless of the trigger
mode.
- [#1596](FuelLabs/fuel-core#1596): Make
`Consensus` type a version-able enum
- [#1593](FuelLabs/fuel-core#1593): Make `Block`
type a version-able enum
- [#1576](FuelLabs/fuel-core#1576): The change
moves the implementation of the storage traits for required tables from
`fuel-core` to `fuel-core-storage` crate. The change also adds a more
flexible configuration of the encoding/decoding per the table and allows
the implementation of specific behaviors for the table in a much easier
way. It unifies the encoding between database, SMTs, and iteration,
preventing mismatching bytes representation on the Rust type system
level. Plus, it increases the re-usage of the code by applying the same
blueprint to other tables.
    
It is a breaking PR because it changes database encoding/decoding for
some tables.
    
    ### StructuredStorage
    
The change adds a new type `StructuredStorage`. It is a wrapper around
the key-value storage that implements the storage
traits(`StorageInspect`, `StorageMutate`, `StorageRead`, etc) for the
tables with blueprint. This blueprint works in tandem with the
`TableWithBlueprint` trait. The table may implement `TableWithBlueprint`
specifying the blueprint, as an example:
    
    ```rust
    impl TableWithBlueprint for ContractsRawCode {
        type Blueprint = Plain<Raw, Raw>;
    
        fn column() -> Column {
            Column::ContractsRawCode
        }
    }
    ```
    
It is a definition of the blueprint for the `ContractsRawCode` table. It
has a plain blueprint meaning it simply encodes/decodes bytes and
stores/loads them into/from the storage. As a key codec and value codec,
it uses a `Raw` encoding/decoding that simplifies writing bytes and
loads them back into the memory without applying any serialization or
deserialization algorithm.
    
If the table implements `TableWithBlueprint` and the selected codec
satisfies all blueprint requirements, the corresponding storage traits
for that table are implemented on the `StructuredStorage` type.
    
    ### Codecs
    
Each blueprint allows customizing the key and value codecs. It allows
the use of different codecs for different tables, taking into account
the complexity and weight of the data and providing a way of more
optimal implementation.
    
That property may be very useful to perform migration in a more easier
way. Plus, it also can be a `no_std` migration potentially allowing its
fraud proving.
    
    An example of migration:
    
    ```rust
    /// Define the table for V1 value encoding/decoding.
    impl TableWithBlueprint for ContractsRawCodeV1 {
        type Blueprint = Plain<Raw, Raw>;
    
        fn column() -> Column {
            Column::ContractsRawCode
        }
    }
    
    /// Define the table for V2 value encoding/decoding.
    /// It uses `Postcard` codec for the value instead of `Raw` codec.
    ///
    /// # Dev-note: The columns is the same.
    impl TableWithBlueprint for ContractsRawCodeV2 {
        type Blueprint = Plain<Raw, Postcard>;
    
        fn column() -> Column {
            Column::ContractsRawCode
        }
    }
    
    fn migration(storage: &mut Database) {
        let mut iter = storage.iter_all::<ContractsRawCodeV1>(None);
        while let Ok((key, value)) = iter.next() {
            // Insert into the same table but with another codec.
            storage.storage::<ContractsRawCodeV2>().insert(key, value);
        }
    }
    ```
    
    ### Structures
    
The blueprint of the table defines its behavior. As an example, a
`Plain` blueprint simply encodes/decodes bytes and stores/loads them
into/from the storage. The `SMT` blueprint builds a sparse merkle tree
on top of the key-value pairs.
    
Implementing a blueprint one time, we can apply it to any table
satisfying the requirements of this blueprint. It increases the re-usage
of the code and minimizes duplication.
    
It can be useful if we decide to create global roots for all required
tables that are used in fraud proving.
    
    ```rust
    impl TableWithBlueprint for SpentMessages {
        type Blueprint = Plain<Raw, Postcard>;
    
        fn column() -> Column {
            Column::SpentMessages
        }
    }
                     |
                     |
                    \|/
    
    impl TableWithBlueprint for SpentMessages {
        type Blueprint =
Sparse<Raw, Postcard, SpentMessagesMerkleMetadata,
SpentMessagesMerkleNodes>;
    
        fn column() -> Column {
            Column::SpentMessages
        }
    }
    ```
    
    ### Side changes
    
    #### `iter_all`
The `iter_all` functionality now accepts the table instead of `K` and
`V` generics. It is done to use the correct codec during
deserialization. Also, the table definition provides the column.
    
    #### Duplicated unit tests
    
The `fuel-core-storage` crate provides macros that generate unit tests.
Almost all tables had the same test like `get`, `insert`, `remove`,
`exist`. All duplicated tests were moved to macros. The unique one still
stays at the same place where it was before.
    
    #### `StorageBatchMutate`
    
Added a new `StorageBatchMutate` trait that we can move to
`fuel-storage` crate later. It allows batch operations on the storage.
It may be more performant in some cases.

- [#1573](FuelLabs/fuel-core#1573): Remove
nested p2p request/response encoding. Only breaks p2p networking
compatibility with older fuel-core versions, but is otherwise fully
internal.

## What's Changed
* Fix the example tx in README.md by @Dentosal in
FuelLabs/fuel-core#1543
* Weekly `cargo update` by @github-actions in
FuelLabs/fuel-core#1557
* Update LICENSE by @Voxelot in
FuelLabs/fuel-core#1558
* Remove duplicating logic in the `KeyValueStore` trait by @xgreenx in
FuelLabs/fuel-core#1559
* Move `KeyValueStore` to the `fuel-core-storage` crate by @xgreenx in
FuelLabs/fuel-core#1566
* Update `libp2p` from `0.50.0` to `0.53.1` by @MitchTurner in
FuelLabs/fuel-core#1379
* Weekly `cargo update` by @github-actions in
FuelLabs/fuel-core#1575
* Weekly `cargo update` by @github-actions in
FuelLabs/fuel-core#1578
* Moved insertion of the blocks into the `BlockImporter` instead of the
executor by @xgreenx in FuelLabs/fuel-core#1577
* Let `NetworkBehaviour` macro generate `FuelBehaviorEvent` in p2p by
@MitchTurner in FuelLabs/fuel-core#1585
* Remove RustSec Ignore for RUSTSEC-2022-0093 by @MitchTurner in
FuelLabs/fuel-core#1586
* Simplify p2p request/response message serialization by @Dentosal in
FuelLabs/fuel-core#1573
* Weekly `cargo update` by @github-actions in
FuelLabs/fuel-core#1588
* Depend on libp2p directly instead of submodules by @MitchTurner in
FuelLabs/fuel-core#1591
* Make `Block` versionable by @MitchTurner in
FuelLabs/fuel-core#1593
* Switch katyo/publish-crates@v2 to xgreenx/publish-crates@v1 for the CI
by @Dentosal in FuelLabs/fuel-core#1603
* Weekly `cargo update` by @github-actions in
FuelLabs/fuel-core#1599
* Fix cargo doc, check for cargo doc in the CI by @Dentosal in
FuelLabs/fuel-core#1601
* Move storage traits implementation to the `fuel-core-storage` crate by
@xgreenx in FuelLabs/fuel-core#1576
* Extract off chain logic from the executor by @xgreenx in
FuelLabs/fuel-core#1579
* Applying comments from the PR for the storage crate by @xgreenx in
FuelLabs/fuel-core#1610
* Use `BlockHeight` as a primary key for the `FuelsBlock` table by
@xgreenx in FuelLabs/fuel-core#1587
* Use `AtomicView` in the `TxPool` by @xgreenx in
FuelLabs/fuel-core#1590
* Unify namespacing in P2P codebase by @MitchTurner in
FuelLabs/fuel-core#1597
* Weekly `cargo update` by @github-actions in
FuelLabs/fuel-core#1611
* Make `Consensus` version-able by @MitchTurner in
FuelLabs/fuel-core#1596
* feat: add api endpoint to retrieve messages by nonce by @matt-user in
FuelLabs/fuel-core#1613
* Update Docker tags for Flux integration by @liviudm in
FuelLabs/fuel-core#1617
* Remove dev consensus key by @MujkicA in
FuelLabs/fuel-core#1614
* Make `BlockHeader` a versionable enum by @MitchTurner in
FuelLabs/fuel-core#1616
* feat: Versionable CompressedCoin by @bvrooman in
FuelLabs/fuel-core#1628
* Use `AtomicView` in all services by @xgreenx in
FuelLabs/fuel-core#1612
* Notify services about importing of the genesis block by @xgreenx in
FuelLabs/fuel-core#1633
* Weekly `cargo update` by @github-actions in
FuelLabs/fuel-core#1634
* feat: Versionable `Message` by @bvrooman in
FuelLabs/fuel-core#1632
* Making relayer independent from the executor by @xgreenx in
FuelLabs/fuel-core#1625
* Upgrade to the fuel-vm 0.44.0 by @Dentosal in
FuelLabs/fuel-core#1600
* Upgrade to fuel-vm 0.45.0 by @MitchTurner in
FuelLabs/fuel-core#1640
* feat: Versionable Merkle metadata by @bvrooman in
FuelLabs/fuel-core#1639
* Add docs for GraphQL DAP endpoints by @Dentosal in
FuelLabs/fuel-core#1636
* Use a separate database for each data domain by @xgreenx in
FuelLabs/fuel-core#1629
* feat: dry run multiple transactions by @matt-user in
FuelLabs/fuel-core#1631
* Weekly `cargo update` by @github-actions in
FuelLabs/fuel-core#1644
* Decrease peer reputation on request timeouts and decode errors by
@Dentosal in FuelLabs/fuel-core#1574
* refactor: remove redudant receipts queries by @matt-user in
FuelLabs/fuel-core#1646
* Latest gas price endpoint by @MitchTurner in
FuelLabs/fuel-core#1649
* Estimate gas price API Endpoint by @MitchTurner in
FuelLabs/fuel-core#1650
* Fix for the race condition with tx status and receipts by @xgreenx in
FuelLabs/fuel-core#1658
* Bump up rust to 1.75 by @MitchTurner in
FuelLabs/fuel-core#1656
* Weekly `cargo update` by @github-actions in
FuelLabs/fuel-core#1655
* Duplication of the `0.22.1` release by @xgreenx in
FuelLabs/fuel-core#1665
* Move updating of the owned messages and coins to off-chain worker by
@xgreenx in FuelLabs/fuel-core#1635
* Weekly `cargo update` by @github-actions in
FuelLabs/fuel-core#1682
* Remove nested contract resolvers in GQL schema by @Voxelot in
FuelLabs/fuel-core#1675
* Soften txpool p2p reputation requirements by @Voxelot in
FuelLabs/fuel-core#1663
* Remove failing deployment job by @Voxelot in
FuelLabs/fuel-core#1692
* chore: format cargo tomls and add ci check by @segfault-magnet in
FuelLabs/fuel-core#1691
* Extraction of the `FuelBlockSecondaryKeyBlockHeights` table to an
off-chain worker by @xgreenx in
FuelLabs/fuel-core#1671
* Weekly `cargo update` by @github-actions in
FuelLabs/fuel-core#1707
* Increase liveness probe timeout to 5 minutes by @rfuelsh in
FuelLabs/fuel-core#1709
* feat: Add `ContractInfo` table and versionable `ContractInfoType` to
on-chain storage by @bvrooman in
FuelLabs/fuel-core#1657
* feat: Versionable `ContractUtxoInfo` for the `ContractsLatestUtxo`
table by @bvrooman in FuelLabs/fuel-core#1712
* Weekly `cargo update` by @github-actions in
FuelLabs/fuel-core#1719
* Bump `fuel-vm` dep to 0.47.1 by @MitchTurner in
FuelLabs/fuel-core#1714
* Bugfix: Calling of the contract with enabled `utxo_validation` fails
by @bvrooman in FuelLabs/fuel-core#1717
* Bugfix: Zero `predicate_gas_used` field during validation of the
produced block by @xgreenx in
FuelLabs/fuel-core#1722
* Notify about imported blocks from the off-chain worker by @xgreenx in
FuelLabs/fuel-core#1723
* Added versioning suffix to the GraphQL endpoint by @xgreenx in
FuelLabs/fuel-core#1725

## New Contributors
* @liviudm made their first contribution in
FuelLabs/fuel-core#1617

**Full Changelog**:
FuelLabs/fuel-core@v0.22.0...v0.23.0
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

Move off-chain database updates from storage trait to off-chain worker
4 participants